From symptom to evidence

Start with a question you can test: which users, service or vantage saw a change, and during which interval? Keep that scope as you move between screens.

Fix the investigation window

Open the incident or test results and select the relevant time window. Distinguish the latest measurement from the sample nearest a selected instant. A screenshot captured after recovery cannot explain what users saw during the failure.

Inspect the returned measurement

Use test results to compare affected vantages. Open record details to inspect the original fields, timestamp and measurement context. Copy or download the returned JSON when preparing a technical handoff. Treat exports as operational data and share only what the recipient needs.

For browser checks, compare timing with the captured screenshot. The Actual size view lets you inspect details without stretching the outer page.

Correlate the machine and its sources

Open infrastructure and inspect the machine’s available views. Host resources, interfaces, containers, observed flows and kernel measurements answer different questions. Check the source and sample time before combining observations.

Follow Open reporting agent to inspect the collector’s configuration, spool and logs. The reporting lease describes reserved reporting activity; it is not a test of target connectivity.

Check the limits of the evidence

A silent hop does not by itself prove forwarded traffic was lost. Read the confidence model and the sample’s recorded method. Missing data may reflect collection, delivery, permissions or the selected window. Use portal troubleshooting to distinguish those cases.

Confirm recovery and hand off

Return to the same vantage and check the same signal after the intervention. Record the time window, affected target, evidence, action taken and what recovered. A correlation is a lead to investigate, not proof of cause.

Use the incident workflow to preserve the operational context. If the findings reveal missing coverage, return to the deployment journey before expanding the rollout.